Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

The first one was a shock. Then came the second. Then a third, each one a different rendering of a creature he had long forgotten. A creature with horns and a mischievous twinkle in the eye. A creature that only existed in his own mind, when he was a child. Now, this creature was in his cousin’s artwork.

He sat on the small chair, staring at the drawing on the floor. His palms were damp. He felt the blood pounding in his ears. He was usually so easy-going, so relaxed. Suddenly, the world felt too small, too close. He wanted to go outside.

"What is this?" he asked his cousin, his voice sounding hollow.

His cousin smiled. “Oh, that’s Gigglesnort! He's my best friend."

Emotion: puzzled

Cluster: Surprise / Confusion
PC1 (Valence): -1.17 Negative
PC2 (Disposition): -0.25

Role in Research

This story is one of 1,000 stories generated for the emotion puzzled. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 puzzled stories, after denoising with neutral dialogue baselines, produces the puzzled emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the puzzled vector is projected through the unembedding matrix.

Promoted:
que0.453
🤔0.421
🤔0.402
🧐0.328
a0.294
Suppressed:
ecstatic-0.265
gladly-0.238
swiftly-0.237
alleviate-0.234
wholeheartedly-0.232